Testing the Limits of AI Scientists
![The [latex]
iceTable[/latex] environment from the [latex]
iceMatrix[/latex] package enables clear table presentation by eschewing vertical separators and employing judicious bolding via the [latex]m[/latex] and [latex]
m[/latex] commands, alongside customizable cell backgrounds.](https://arxiv.org/html/2602.06855v1/x2.png)
A new benchmark suite challenges artificial intelligence to perform core tasks within the scientific research process.
Science
Smarter Robot Movement: Blending Precision and Speed
A new framework combines the strengths of both optimization-based and analytic inverse kinematics to achieve more efficient and robust robotic motion planning.
AI Designs High-Efficiency Solar Cells, Closing the Loop from Research to Reality
![The PVK-LLM framework advances scientific discovery through an iterative, closed-loop active learning process, leveraging a knowledge base constructed from scientific literature to refine a language model across three stages-domain knowledge acquisition ([latex]Stage \, I[/latex]), knowledge grounding with experimental data ([latex]Stage \, II[/latex]), and structured knowledge graph creation ([latex]Stage \, III[/latex])-thereby enabling autonomous design, verification, and optimization cycles within both simulated and laboratory environments, as demonstrated by a corpus of perovskite literature spanning from 2018 to 2025.](https://arxiv.org/html/2602.04914v1/x1.png)
A new approach combines the power of artificial intelligence with materials science to autonomously develop perovskite solar cell compositions, accelerating the pace of innovation in renewable energy.
Dancing Droplets: Unveiling Complex Dynamics in Soft Landscapes
![A particle navigating a non-smooth harmonic potential experiences a wave-memory force generated by its own motion, effectively creating a self-interaction alongside a drag force, where the wave’s spatial form is described by [latex]W(x)=\cos(x)[/latex] and decays exponentially over time.](https://arxiv.org/html/2602.05913v1/FIG1.png)
A new study reveals how walking droplets exhibit surprisingly rich and chaotic behavior when confined by textured surfaces and driven by internal wave effects.
Orchestrating AI: Kubernetes Powers the Next Wave of Inference
A new study details how Kubernetes, enhanced with open-source tools, can efficiently manage the demanding workloads of modern generative AI applications.
Beyond Experience: Guiding Agents to Adapt and Improve
A new method allows AI agents to intelligently leverage past successes in novel situations by subtly adjusting their internal understanding of the world.
Disorderly Systems, Linear Insights
![A universal linear structure-property relationship (SPR) is established for predicting phonon-related properties of amorphous materials, where atomic configurations are encoded via a symmetry-invariant radial distribution function [latex]RDF_g(r)g(r)[/latex] and mapped to target observables-including phonon density of states, Raman, inelastic neutron scattering, and inelastic X-ray scattering spectra-through learnable weights and biases shared across diverse amorphous systems, encompassing monolayers, bulk glasses, and high-entropy alloys with varying degrees of structural disorder.](https://arxiv.org/html/2602.05313v1/x1.png)
New research reveals a surprisingly simple relationship between structure and vibrational properties in amorphous materials, challenging the need for complex modeling.
Spreading Out Smarter: A New Challenge for Robot Swarms
![The study establishes bounds for multi-robot dispersion, demonstrating that universal exploration sequences require memory [latex]M^*[/latex] dependent on the least inter-robot distance [latex]iiis[/latex], the number of nodes [latex]n~(k)[/latex], and maximum degree Δ, with performance varying based on prior knowledge of these parameters-or the lack thereof-and where computational complexity is expressed as time [latex]T[/latex] and memory [latex]M[/latex] with polylogarithmic factors suppressed by [latex]\tilde{O}[/latex].](https://arxiv.org/html/2602.05948v1/x9.png)
Researchers have formalized a more complex version of the classic ‘dispersion’ problem, requiring robots to navigate graphs and find nodes matching their assigned color, demonstrating increased difficulty over previous models.
Beyond Bedside Manner: How Medical Students Want AI Patients to Learn
New research reveals that medical learners prioritize effective instruction and focused practice over realistic conversation when co-designing AI-powered standardized patients.
Seeing People in 3D: A Sensor Showdown
![The research visualizes common data corruptions inherent in multi-sensor perception systems, specifically demonstrating how noise affects both camera imagery and [latex] 360^\circ [/latex] LiDAR point clouds- modalities labeled as C and L respectively-as captured within the JRDB dataset.](https://arxiv.org/html/2602.05538v1/x7.png)
A new study rigorously compares how different sensor technologies – cameras, LiDAR, and their combinations – perform in detecting people across real-world indoor and outdoor environments.